Merge pull request #1 from chechu/master

Conjugaciones

software/build.gradle

@@ -0,0 +1,26 @@
+apply plugin: 'groovy'
+apply plugin: 'application'
+apply plugin: 'maven'
+
+group = 'com.acceso'
+version = '0.1-SNAPSHOT'
+
+/*
+ * Mapeando la configuración 'groovy' de Gradle con el scope 'runtime' de maven. De esta forma
+ * al generar el pom considerará como dependencias a incluir aquellas asociadas con la configuración
+ * 'groovy'. Por defecto considera únicamente las que incorpora el plugin 'java' de gradle.
+ */
+conf2ScopeMappings.addMapping(MavenPlugin.COMPILE_PRIORITY, configurations.groovy, org.gradle.api.artifacts.maven.Conf2ScopeMappingContainer.COMPILE)
+
+repositories {
+	mavenLocal()
+	mavenCentral()
+}
+
+dependencies {
+	groovy group: 'org.codehaus.groovy', name: 'groovy-all', version: '1.8.6'
+	groovy group: 'log4j', name: 'log4j', version: '1.2.16'
+	groovy 'org.jsoup:jsoup:1.6.3'
+}
+
+mainClassName = 'com.acceso.lemarios.ExtractInflectedForms'

software/settings.gradle

@@ -0,0 +1 @@
+rootProject.name = 'lemarios'

software/src/main/groovy/com/acceso/lemarios/ExtracInflectedForms.groovy

@@ -0,0 +1,102 @@
+package com.acceso.lemarios
+
+import org.jsoup.nodes.Document
+import org.jsoup.Jsoup
+import org.jsoup.select.Elements
+
+class ExtractInflectedForms {
+
+	private final static String OUTPUT_FILE = '/tmp/inflectedForms.txt'
+	private final static String VERB_FILE = '/com/acceso/lemarios/verbos-espanol.txt'
+	private final static String LEMA_SEARCH_URL = 'http://lema.rae.es/drae/srv/search?val='
+	private final static String INFLECTED_FORMS_URL = 'http://lema.rae.es/drae/srv/'
+	private final static List STOP_WORDS = ['me', 'te', 'se', 'nos', 'os', 'etc.'] 
+
+	private File outputFile = new File(OUTPUT_FILE)
+
+	public void execute() {
+		List<String> inflected = []
+		List<String> verbs = loadVerbs()
+		verbs.eachWithIndex { verb, index ->
+			println "[${index}, ${verbs.size()}]"
+			inflected.addAll(extract(verb))
+		}
+
+		//inflected = inflected.sort().unique()
+		inflected.removeAll(STOP_WORDS)
+
+		save(inflected)
+	}
+
+	private List<String> extract(String verb) {
+		List<String> results = []
+		List<String> inflectedUrls = getInflectedUrls(verb)
+
+		//println "[${verb}] (${inflectedUrls.size()})"
+		//inflectedUrls.each { println "\t${it}" }
+
+		results = inflectedUrls.collect { 
+			try {
+				return getInflectedForms(verb, it)
+			} catch(Exception e) {
+				println "Error [${verb}] [${it}]: ${e}"
+				return []
+			}
+		}.flatten()
+
+		//println "Declinaciones " + results.join('--')
+
+		return results
+	}
+
+	private List<String> getInflectedForms(String verb, String inflectedUrl) {
+		Document doc = Jsoup.connect(inflectedUrl).get()
+		Elements elements = doc.select("p[class=z]")
+
+		List<String> results = elements.collect { element ->
+			element.text().split(',|/| o | u | ').collect{it.trim()}.findAll{it}
+		}.flatten()
+
+		if(!results) {
+			println "[0 inflected forms for ${verb}]"
+		}
+
+		return results
+	}
+
+	// Devuelve las urls con las declinaciones
+	private List<String> getInflectedUrls(String verb) {
+		List<String> result = []
+
+		try {
+			Document doc = Jsoup.connect(LEMA_SEARCH_URL + URLEncoder.encode(verb, 'ISO-8859-1')).get()
+			Elements elements = doc.select("a img[alt^=Ver conjugaci]")
+
+			result = elements.collect { element -> INFLECTED_FORMS_URL + element.parent().attr('href') }
+		} catch (Exception e) {
+			println "Error [${verb}] [${it}]: ${e}"
+		}
+
+		return result
+	}
+
+	private void save(List<String> forms) {
+		forms.each { outputFile << "${it}\n" }
+	}
+
+	private List<String> loadVerbs() {
+		List<String> result = []
+		InputStream i = getClass().getResourceAsStream(VERB_FILE)
+
+		if(i) {
+			i.eachLine { line -> if(line.contains('í')) result << line }
+		}
+
+		return result
+	}
+
+	public static void main(String[] args) {
+		ExtractInflectedForms e = new ExtractInflectedForms()
+		e.execute()
+	}
+}